On Thursday, June 24, 2004, 10:50:34 AM, Herb wrote:

HG> In the last couple days we are seeing quite an increase in the amount of
HG> spam leaking past sniffer and declude.  Anyone else seeing this?

We have quite a spam storm going...
I personally had a 4 hour rule coding session last night - all new
rules, hundreds, and hundreds of new spam loosed on the world. I think
we're getting through this pulse though.

See:

http://www.sortmonster.com/MessageSniffer/Performance/ChangeRates.jsp

Seems almost like they were taking the weekend off (see day 4,5,6)

Days Ago Adjustments
-------- -----------

0        540
1        409
2        344
3        411
4        169
5        210
6        385
7        853
8        341
9        365
10       302
11       177
12       299
13       207
14       680

Day 2 showed nominally, but since then 409 + 540.
The 540 for today is just what's been coded so far - it's beginning to
look like we're headed for another ~1000 rule day.

There seems to be a recent pattern emerging where there are large
spikes followed by relatively lower, more stable numbers. Note the
spike in progress today (day 0), back on day 7, and again on day 14.
Some of this is clearly associated with weekends and a lack of reports
from user submissions - but certainly not all of it.

We are working to expand our spamtrap network so that our training
inputs can be more consistent. Anyone with clean spamtraps that we are
not yet seeing should let us know at support@ if they are willing to
share. We can set up special addresses for collecting this data and
reviewing it. (We treat spamtraps differently than submissions to
spam@)

Another thing I've noted is that the ROKSO folks are on the move...
that is, the heavy spamers listed in ROKSO _seem_ to be on a run to
light up new IPs - especially in smaller blocks (/25, /28, etc...)
This isn't usually what we see - It makes me think that some of them
have decided to try some new things and we're just not in the loop yet
as to what those new things are.

It's mostly intuition because the data is still to fuzzy, but
something is going on.

I look forward to any other thoughts folks may have.

My $0.02
_M




This E-Mail came from the Message Sniffer mailing list. For information and 
(un)subscription instructions go to 
http://www.sortmonster.com/MessageSniffer/Help/Help.html

Reply via email to